Measurement Framework and Case Study Trials of the ......IRIM2019 –25-27 February 2019, Kathmandu, Nepal •100 x 100 meter grid cells representing population derived from census

IRIM2019 – 25-27 February 2019, Kathmandu, Nepal

Measurement Framework and Case

Study Trials of the Proposed Methodology

Kevin McPherson


Malawi


• 100 x 100 meter grid cells representing population derived from census data and imagery

• Reconciled to the lowest level of country census data (e.g. ward)• WorldPop are making annualised data available for last 20 years

World Pop data


“The distinction between urban and rural population is not amenable to a single definition applicable to all countries.

For this reason, each country should decide which areas are to be classified as urban and which as rural, in accordance with their own circumstances”.

UN Statistics Division website

Urban / rural definition



Malawi Fourth Integrated Household Survey 2016/17:

“The urban strata include the four major urban areas: Lilongwe City, Blantyre City, Mzuzu City, and the Municipality of Zomba.

All other areas are considered as rural areas, and each of the 27 districts were considered as a separate sub-stratum as part of the main rural stratum”.



Malawi Urbanization Review (GSURR, 2015)



“The figure may be different from the official statistics because a different definition of urban areas is used.”

“Urban areas are defined based on the 1995 University of Columbia (CIESIN) urban area imagery”.

WB RAI Report, 2018 (draft)


Urban / rural boundary

Source Year Population

Blantyre City WorldPop 2015 884,817

Malawi NSO (Malawi in Figures, 2016) 2015 884,000

Lilongwe City WorldPop 2015 922,374

Malawi NSO (Malawi in Figures, 2016) 2015 1,037,000


Urban / rural boundary

Humanitarian Data Exchange Level 3 Boundaries


SourceLength of Network

(km)

Malawi Road Network Classification (2008) 15,457

Open Street Map (2018) 15,540

WB RAI Report (2016) 12,859

GRIP Database (2018) tbc

Road Network Lengths

These figures represent the Classified Road Network only


OSM v GRIP


OSM v GRIP


• Only 13.6% of records have any surface information

• Only 966 km of the entire network is identified as “paved”, versus the 4,100 km on the MRA website

• Only 2,018 records (0.8%) have “smoothness” information

• Only 653 records (0.3%) have “seasonal” information

• Only 289 records have (0.1%) max speed information

OSM Data Breakdown


Alternative Methodologies

• Mobile Phone Network Data

• “Machine Learning” – e.g.

o to produce mapping of Road Networks from imagery

o to determine if road is “all-season” or not (e.g. based on flooding that may be visible on historic satellite imagery)


Alternative Methodologies

• Mobile Phone Network Data


Alternative Methodologies• Machine Learning Tools



Key Points for a New Methodology

• Clear assessment of data sources identified and chosen• Clear documentation on data sources and methodology• Validation / assurance on the data sources and

methods• Consistency over time in the method of calculation to

allow for trend analysis• Consistency of urban/rural definition among countries

(UN is working on this*)• NSOs must always play a major role in definition /

coordination / assurance / publication process

www.research4cap.org

Follow ReCAP on:

#IRIM2019

Thank you for your attention


Documents

Measurement Framework and Case Study Trials of the ......IRIM2019 –25-27 February 2019, Kathmandu, Nepal •100 x 100 meter grid cells representing population derived from census